News
The raw AI news stream — everything we ingest across RSS, Hacker News, arXiv, GitHub, Bluesky, YouTube and more, newest first.
| Source | Age | Title | Summary | Status |
|---|---|---|---|---|
hn hn-ai | 17d | Ask HN: How do small teams pragmatically manage Dependabot fatigue? | Amount of CVEs nowadays is kind of madness and I'm wondering how you're all dealing with the updates? I know larger organiza… | ↗Not briefed |
arxiv arxiv-ai | 17d | Different Corpora, Different Skills? A Paired Study of Small Language Models | Small language models are often trained under tight data budgets, where the choice of pretraining corpus may matter as much as model s… | ↗Not briefed |
hn hn-ai | 17d | Show HN: teletap – Your remote coding agents ask you when stuck | Hi HN!<p>I'm Manuel, CTO of Tandem.<p>Most of our PRs are now written by custom coding agents on remote hosts. But we had a recur… | ↗Not briefed |
hn hn-ai | 17d | Show HN: Vector – a macOS app for engineering with agents | Vector is a MacOS app for orchestrating agents. It's similar to T3 code and the features are basic, but I have some things to try… | ↗Not briefed |
hn hn-ai | 17d | KVGuard: Catching the KV cache bug behind Cloudflare's GLM leak | — | ↗Not briefed |
arxiv arxiv-ai | 17d | Decodability Is Not Validity: Evaluating Nuisance Sensitivity in Language Model Concept Probes | Two concept probes can be equally accurate at predicting a concept from a language model's hidden states while their scores contain di… | ↗Not briefed |
bluesky bluesky-ai | 17d | OpenAI pledges to change how it handles data center deals, including avoiding nondisclosure agreements and cutting power usage, after Democ… | OpenAI pledges to change how it handles data center deals, including avoiding nondisclosure agreements and cutting power usage, after… | ↗Not briefed |
rss techmeme | 17d | OpenAI pledges to change how it handles data center deals, including avoiding nondisclosure agreements and cutting power usage, after Democrats' midterm wins (Dina Bass/Bloomberg) | Dina Bass / Bloomberg: OpenAI pledges to change how it handles data center deals, including avoiding nondisclosure agreements and cutt… | ↗Not briefed |
rss theverge-ai | 17d | Big Tech’s big AI buildout will not end until someone stops it | After four of the biggest competitors in the AI race held quarterly earnings calls late last month, two things are clear: Big Tech is… | ↗Not briefed |
rss techmeme | 17d | Sunil Mittal, founder of India's second-largest telco Airtel, says Airtel-backed OneWeb can build a satellite-internet network independent of the US and China (Aryan Gupta/Bloomberg) | Aryan Gupta / Bloomberg: Sunil Mittal, founder of India's second-largest telco Airtel, says Airtel-backed OneWeb can build a satellite… | ↗Not briefed |
rss bing-news-ai | 17d | New corridor eyed for 'super intelligence' projects | MANILA, Philippines — The Philippines plans to introduce a technology corridor to attract investments across the “super intelligence”… | ↗Not briefed |
arxiv arxiv-ai | 17d | Dataset Collection and Fine-Tuning of NLLB for German--English Translation | This paper describes the curation of a high-quality parallel corpus from OPUS Books and the fine-tuning of Meta's NLLB-200 models for… | ↗Not briefed |
bluesky bluesky-ai | 17d | OpenAI-CEO Sam Altman legte vergangene Woche nahe, die weltweiten RAM-Vorräte besser in KI-Rechenzentren zu stecken. Nun macht er eine Roll… | OpenAI-CEO Sam Altman legte vergangene Woche nahe, die weltweiten RAM-Vorräte besser in KI-Rechenzentren zu stecken. Nun macht er eine… | ↗Not briefed |
bluesky bluesky-ai | 17d | I THINK NOT, STEAM. I THINK NOT. | I THINK NOT, STEAM. I THINK NOT. | ↗Not briefed |
rss techmeme | 17d | Sony reports Q2 revenue of ~$21.7B, operating profit of $3.94B, both above est., and 2.6M PS5 sales, as GTA VI sparks a US sales jump, especially of the PS5 Pro (Takashi Mochizuki/Bloomberg) | Takashi Mochizuki / Bloomberg: Sony reports Q2 revenue of ~$21.7B, operating profit of $3.94B, both above est., and 2.6M PS5 sales, as… | ↗Not briefed |
bluesky bluesky-ai | 17d | 引き継ぎでSEと話していて、なんかフワフワしてると思ったら、データの解析は全部LLM任せで、内容を全く把握していなかったので、そりゃフワフワするよな、と納得した。それでいて成果物はちゃんと?仕上げるようなシステムになっていたから、何だか不思議な気分になったよ。 魔法の箱に近付い… | 引き継ぎでSEと話していて、なんかフワフワしてると思ったら、データの解析は全部LLM任せで、内容を全く把握していなかったので、そりゃフワフワするよな、と納得した。それでいて成果物はちゃんと?仕上げるようなシステムになっていたから、何だか不思議な気分になったよ。 魔法の… | ↗Not briefed |
arxiv arxiv-ai | 17d | EnergyBench: a live, gate-aligned benchmark of day-ahead forecasts for European power markets | Energy forecasting models are usually compared on backtests and fixed public datasets. Backtests can use data that did not yet exist w… | ↗Not briefed |
hn hn-ai | 17d | Show HN: Distill anyone's writing style into a skill | — | ↗Not briefed |
hn hn-ai | 17d | KVGuard: A KV Cache Safety Checker for LLM Serving Systems | — | ↗Not briefed |
hn hn-ai | 17d | Kinference: Open-source inference control plane for vLLM🔧 vLLM | — | ↗Not briefed |
arxiv arxiv-ai | 17d | PBPO: Prefix-Block Policy Optimization for Diffusion Language Models | Diffusion Large Language Models (dLLMs) commonly decode responses by proceeding autoregressively across blocks while iteratively denoi… | ↗Not briefed |
bluesky bluesky-ai | 17d | 今のSEって当たり前のようにLLM使うよなぁ。 | 今のSEって当たり前のようにLLM使うよなぁ。 | ↗Not briefed |
arxiv arxiv-ai | 17d | Bridging Structure and Semantics: Knowledge Graph Verification via Triple-to-Text Alignment | Semantic accuracy is an important dimension of knowledge graph quality. Graph-based scoring methods exploit relational structure, whil… | ↗Not briefed |
bluesky bluesky-ai | 17d | To keep your notes safe, we've completed a full security review of Joplin Server with help from Anthropic, and fixed 50+ issues. If you se… | To keep your notes safe, we've completed a full security review of Joplin Server with help from Anthropic, and fixed 50+ issues. If y… | ↗Not briefed |
arxiv arxiv-ai | 17d | The Intrusion Pathway: How Non-Evidential Context Degrades Language Model Performance | Language models lose accuracy when a prompt contains non-evidential context: irrelevant statements from unrelated domains, or random o… | ↗Not briefed |
bluesky bluesky-ai | 17d | おれは資本主義の犬.... LLMの奴隷... | おれは資本主義の犬.... LLMの奴隷... | ↗Not briefed |
arxiv arxiv-ai | 17d | Adapters as Data: The Latent Space of LoRAs | LoRA adapters are usually treated as finished artifacts rather than as a data modality. Prior work has shown that LoRA adapters popula… | ↗Not briefed |
arxiv arxiv-ai | 17d | Optimal Learning in Games with Arbitrary Delays: Best-of-Both-Worlds for Self-Play and Adversarial Opponents | We study learning in zero-sum games with delayed feedback. We introduce Multi-Anchor Optimistic Online Mirror Descent (MAO-OMD), a sin… | ↗Not briefed |
bluesky bluesky-ai | 17d | OpenAI pledges to pay for data center power after election highlights AI concerns | OpenAI pledges to pay for data center power after election highlights AI concerns | ↗Not briefed |
hn hn-ai | 17d | Craft Anything: understand how any software works, and craft your own | — | ↗Not briefed |
arxiv arxiv-ai | 17d | RIVER: Evidence-guided multi-agent retrieval for enzyme--reaction discovery | Enzyme function annotation and the identification of candidate enzymes for a given reaction are essential steps in connecting genomic… | ↗Not briefed |
arxiv arxiv-ai | 17d | Scaling Laws for SAE Training Data | Activation storage is often a bottleneck in sparse autoencoder (SAE) training. We study how many unique activations an SAE needs when… | ↗Not briefed |
arxiv arxiv-ai | 17d | Automated Video-to-QA Pipeline: End-to-End Question Generation and Verification Using Large Language Models🔧 Whisper | The exponential growth of educational and informational video content on platforms like YouTube presents a challenge for efficient kno… | ↗Not briefed |
arxiv arxiv-ai | 17d | Neural Collapse on Imbalanced Datasets and Open-Set Scenario | Neural collapse (NC) describes the elegant geometric structure that sample features and classifier vectors exhibit when a classificati… | ↗Not briefed |
bluesky bluesky-ai | 17d | Anthropic? | Anthropic? | ↗Not briefed |
arxiv arxiv-ai | 17d | Foresee to Trade: Forecast-Augmented Policy Optimization for Portfolio Management | Online investment services must translate asset opportunities into investment outcomes for users' accounts, yet the same market opport… | ↗Not briefed |
arxiv arxiv-ai | 17d | ConvPE: Convolutional Position Embedding Improves Transformers | ConvPE (Convolutional Position Embedding) improves both the performance and efficiency of Transformer models in language modeling, lon… | ↗Not briefed |
bluesky bluesky-ai | 17d | For starters we’d all have to be completely fine with the UK answer to Anthropic running roughshod over copyright and intellectual property… | For starters we’d all have to be completely fine with the UK answer to Anthropic running roughshod over copyright and intellectual pro… | ↗Not briefed |
arxiv arxiv-ai | 17d | Ordering Multivariate Time Series by Volatility Memory: A Temporal Convolutional Approach to Volatility Heredity | Volatility dynamics can carry information about temporal relationships that is not apparent in the conditional mean. Extracting this i… | ↗Not briefed |
arxiv arxiv-ai | 17d | Comparing Classification and Feature Learning by Analyzing Their Loss Functions in Normalized Spaces with Different Scale Factors | Classification and feature learning are two closely interrelated fundamental tasks. Feature learning in supervised settings is typical… | ↗Not briefed |
hn hn-ai | 17d | Show HN: Review Lens – review pull requests inside your JetBrains IDE | — | ↗Not briefed |
arxiv arxiv-ai | 17d | Extending Reasoning in Typed Decision Models via Looping | Typed decision models choose among a fixed set of answer options in a single forward pass, so unlike generative language models they c… | ↗Not briefed |
arxiv arxiv-ai | 17d | Evaluating Token Probabilities as Confidence Estimators in Human Behavior Simulations: A Case Study in Luxembourgish Next Like Prediction | As large language models (LLMs) are increasingly used to simulate human behavior, their operational validity is typically assessed thr… | ↗Not briefed |
arxiv arxiv-ai | 17d | O$^2$PD: Overlap-Gated On-Policy Distillation on Reliable Prefix | On-policy distillation (OPD) supplies dense token-level supervision by aligning a student to the teacher's next-token distribution on… | ↗Not briefed |
arxiv arxiv-ai | 17d | Seeing Beyond the Next Step: World-Model-Guided Human-Like Navigation in Multi-Agent Scenes | Selecting a next motion in a populated scene is not just a matter of avoiding what is currently in the way: an action can be locally f… | ↗Not briefed |
bluesky bluesky-ai | 17d | 🤖 Trajectory Secures $45M to Automate AI Model Fine-Tuning Trajectory has closed a $45 million Series A funding round to scale its platfo… | 🤖 Trajectory Secures $45M to Automate AI Model Fine-Tuning Trajectory has closed a $45 million Series A funding round to scale its p… | ↗Not briefed |
arxiv arxiv-ai | 17d | Static Recovery Is Not Dynamic Stability: Dynamics-Aware Benchmarking of Protein Motif Scaffolding | Protein motif scaffolding benchmarks assess whether generative models can build scaffolds around functionally important residues. Thei… | ↗Not briefed |
arxiv arxiv-ai | 17d | Reinforcing What the Student Has Learned: Teacher-Guided On-Policy Distillation | On-policy distillation (OPD) improves language-model reasoning through dense teacher feedback on student-generated responses.We first… | ↗Not briefed |
arxiv arxiv-ai | 17d | Continued Training of Continuous Language Flows with Token Identity Squared Supervision | Diffusion and flow models have recently been extended to language modeling, generating text through continuous denoising in token embe… | ↗Not briefed |
arxiv arxiv-ai | 17d | Evaluating Language Model Features for Zero-Shot Time-Series Anomaly Detection Under Isolation Assumption🔧 Modal | Time-Series Anomaly Detection (TSAD) is a fundamentally challenging task: it is unsupervised, requires high local precision capabiliti… | ↗Not briefed |